Research map / Image generation / Text-to-image diffusion
Diverse and Coherent Paragraph Generation from Images
Moitreya Chatterjee, Alexander G. Schwing
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add coherence vectors (ablation) | Ours · Stanford Image-Paragraph test | CIDEr: 16.42 -> 19.95 | Improved |
| add coherence vectors (ablation) | Ours · Stanford Image-Paragraph test | METEOR: 16.89 -> 17.81 | Improved |
| add coherence vectors (ablation) | Ours · Amazon Office-Products reviews test | CIDEr: 15.39 -> 16.05 | Improved |
| add coherence vectors (ablation) | Ours · Amazon Office-Products reviews test | METEOR: 15.86 -> 16.58 | Improved |
| add global topic vector (ablation) | Ours · Stanford Image-Paragraph test | CIDEr: 18.23 -> 19.95 | Improved |
| add global topic vector (ablation) | Ours · Amazon Office-Products reviews test | CIDEr: 15.54 -> 16.05 | Improved |
| global vector alone vs coherence vectors alone: coherence vectors matter more (ablation) | Ours · Stanford Image-Paragraph test | CIDEr: 16.42 -> 18.23 | Improved |
| train with VAE for diversity | Ours · Stanford Image-Paragraph test | CIDEr: 19.95 -> 20.93 | Improved |
| train with VAE for diversity | Ours · Amazon Office-Products reviews test | CIDEr: 16.05 -> 17.17 | Improved |
| train in GAN setting instead of maximum likelihood | Ours · Stanford Image-Paragraph test | CIDEr: 19.95 -> 18.05 | Hurt |
| train in GAN setting instead of maximum likelihood | Ours · Amazon Office-Products reviews test | CIDEr: 16.05 -> 15.43 | Hurt |
| VAE-trained model vs RTT-GAN (Plus) using extra COCO data | Ours (with VAE) · Stanford Image-Paragraph test | CIDEr: 20.36 -> 20.93 | Improved |
Closest papers by meaning
- Matching Visual Features to Hierarchical Semantic Topics for Image Paragraph Captioning (IJCV 2022 · similarity 0.91)
- A Hierarchical Approach for Generating Descriptive Image Paragraphs (CVPR 2017 · similarity 0.90)
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · similarity 0.86)
- Harmonizing Visual Text Comprehension and Generation (NeurIPS 2024 · similarity 0.86)
- Progressive Text-to-Image Diffusion with Soft Latent Direction (AAAI 2024 · similarity 0.85)
- CAPability: A Comprehensive Visual Caption Benchmark for Evaluating Both Correctness and Thoroughness (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
